We propose longitudinal and microgenetic evaluation of early human vocalizations as a foundation for speech. The work starts with focus on typical vocal development within an evolutionarily-informed, dynamic systems framework illuminating key features of vocal development. The work proceeds to test the key features in terms of their roles in predicting development and anomalies of development (autism, language delay). The basic work focuses on: 1) the infant tendency to actively explore vocalization, creating and systematically repeating categories (at first, squeals, growls, vowel-like sounds, etc.) that become increasingly speech-like across the 1st year, 2) the infant tendency to engage in interaction using those categories; 3) the caregiver tendency to attend to infant vocalization and recognize both the infant vocal categories and how they express infant state and social functions; and 4) the dynamic interaction, where caregiver and infant respond systematically to vocalizations and modulate reactions in ways that are specific to the infant vocal categories and their functions. This work breaks ground by: 1) focusing in much more detail than in prior work on the vocal categories that form the fulcrum of caregiver-infant vocal interaction, affecting both vocalizations and their social functions; 2) improving perspectives on change through microgenetic approaches with computer-interactive diary data from caregivers, extensive laboratory recordings, and all-day recordings in the home; 3) enhancing perspectives on both infant and caregiver vocalizations through sophisticated coding of both lab and home recordings by trained observers and acoustic analysts across many dimensions of action during the 1st 30 months of life; 4) exploiting advanced quantitative tools (lag sequential analysis, recurrence quantification analysis, and neural network modeling) to facilitate rapid and insightful processing of vocalization data; and 5) providing automated analysis of infant vocal categories and caregiver vocal reactions at unprecedented scales (thousands of hours of naturalistic recording) informed and validated by an extensive body of human coded and acoustically analyzed data. These approaches will yield extraordinary potential impact by providing an empirical typically-developing reference point for developmental interpretation. Impact studies will test the key variables of vocal category development and interaction determined by the research using automated analytical tools also produced in the research as predictors of 1) development, 2) differentiation of three groups of children (typically developing, autistic, language delayed), and 3) outcome measures of language and social development in an existing database including 1486 all-day recordings.